HS project

Zoom in immune cells

This file is used to analyse the immune cells dataset.

library(dplyr)
library(patchwork)
library(ggplot2)
library(ComplexHeatmap)
library(org.Mm.eg.db)

.libPaths()

## [1] "/usr/local/lib/R/library"

Preparation

In this section, we set the global settings of the analysis. We will store data there :

save_name = "immune_cells"
out_dir = "."

We load the dataset :

sobj = readRDS(paste0(out_dir, "/", save_name, "_sobj.rds"))
sobj

## An object of class Seurat 
## 15121 features across 2329 samples within 1 assay 
## Active assay: RNA (15121 features, 2000 variable features)
##  6 dimensional reductions calculated: RNA_pca, RNA_pca_20_tsne, RNA_pca_20_umap, harmony, harmony_20_umap, harmony_20_tsne

We load the sample information :

sample_info = readRDS(paste0(out_dir, "/../../1_metadata/hs_hd_sample_info.rds"))
project_names_oi = sample_info$project_name

graphics::pie(rep(1, nrow(sample_info)),
              col = sample_info$color,
              labels = sample_info$project_name)

Here are custom colors for each cell type :

color_markers = readRDS(paste0(out_dir, "/../../1_metadata/hs_hd_color_markers.rds"))

data.frame(cell_type = names(color_markers),
           color = unlist(color_markers)) %>%
  ggplot2::ggplot(., aes(x = cell_type, y = 0, fill = cell_type)) +
  ggplot2::geom_point(pch = 21, size = 5) +
  ggplot2::scale_fill_manual(values = unlist(color_markers), breaks = names(color_markers)) +
  ggplot2::theme_classic() +
  ggplot2::theme(legend.position = "none",
                 axis.line = element_blank(),
                 axis.title = element_blank(),
                 axis.ticks = element_blank(),
                 axis.text.y = element_blank(),
                 axis.text.x = element_text(angle = 30, hjust = 1))

This is the projection of interest :

name2D = "harmony_20_tsne"

We design a custom functions to represent cells of interest on the projection :

see_clusters = function(pop_oi = "DC") {
  # Clusters containing population of interest
  clusters_oi = names(which(cell_type_clusters == pop_oi))
  
  # Colors for clusters
  custom_colors = aquarius::gg_color_hue(length(levels(sobj$seurat_clusters)))
  names(custom_colors) = levels(sobj$seurat_clusters)
  custom_colors[!(names(custom_colors) %in% clusters_oi)] = "gray92"
  
  p1 = Seurat::DimPlot(sobj, reduction = name2D, cols = custom_colors,
                       group.by = "seurat_clusters", label = TRUE) +
    ggplot2::labs(title = "Clusters of interest",
                  subtitle = paste0(clusters_oi, collapse = ", ")) +
    ggplot2::theme(aspect.ratio = 1,
                   plot.title = element_text(hjust = 0.5),
                   plot.subtitle = element_text(hjust = 0.5)) +
    Seurat::NoAxes() + Seurat::NoLegend()
  
  # Color for cell type
  custom_colors = unlist(color_markers)
  custom_colors[!(names(custom_colors) %in% pop_oi)] = "gray92"
  
  p2 = Seurat::DimPlot(sobj, reduction = name2D,
                       group.by = "cell_type", cols = custom_colors) +
    ggplot2::labs(title = "Annotation of interest",
                  subtitle = pop_oi) +
    ggplot2::theme(aspect.ratio = 1,
                   plot.title = element_text(hjust = 0.5),
                   plot.subtitle = element_text(hjust = 0.5)) +
    Seurat::NoAxes() + Seurat::NoLegend()
  
  # Piechart for each cluster, by sample name
  plot_list = lapply(clusters_oi, FUN = function(one_cluster) {
    df = sobj@meta.data %>%
      dplyr::filter(.data$seurat_clusters == .env$one_cluster) %>%
      dplyr::select(sample_identifier, seurat_clusters)
    
    p = aquarius::plot_piechart(df,
                                grouping_var = "sample_identifier",
                                colors = sample_info$color) +
      ggplot2::labs(title = paste0("Cluster ", one_cluster),
                    subtitle = paste0(nrow(df), " cells")) +
      ggplot2::theme(plot.title = element_text(hjust = 0.5),
                     plot.subtitle = element_text(hjust = 0.5))
    
    return(p)
  })
  p3 = patchwork::wrap_plots(plot_list)
  
  # Patchwork
  p = patchwork::wrap_plots(p2, p1, p3, nrow = 1)
  
  return(p)
}

We design a custom function to make the GSEA plot and a word cloud graph :

make_gsea_plot = function(gsea_results, gs_oi, fold_change, metric = "FC") {
  fold_change$metric = fold_change[, metric]
  
  plot_list = lapply(gs_oi, FUN = function(gene_set) {
    # Gene set content
    gs_content = gene_sets %>%
      dplyr::filter(gs_name == gene_set) %>%
      dplyr::pull(ensembl_gene) %>%
      unique()
    
    # Gene set size
    nb_genes = length(gs_content)
    
    # Enrichment metrics
    NES = gsea_results@result[gene_set, "NES"]
    p.adjust = gsea_results@result[gene_set, "p.adjust"] %>%
      round(., 4)
    qvalues = gsea_results@result[gene_set, "qvalues"]
    
    if (p.adjust > 0.05) {
      p.adjust = paste0("<span style='color:red;'>", p.adjust, "</span>")
    }
    
    my_subtitle = paste0("\nNES : ", round(NES, 2),
                         " | padj : ", p.adjust,
                         " | qval : ", round(qvalues, 4),
                         " | set size : ", nb_genes, " genes")
    
    # Size limits
    lower_FC = min(fold_change[gs_content, ]$metric, na.rm = TRUE)
    upper_FC = max(fold_change[gs_content, ]$metric, na.rm = TRUE)
    
    # Plot
    p = enrichplot::gseaplot2(x = gsea_results, geneSetID = gene_set) +
      ggplot2::labs(title = gene_set,
                    subtitle = my_subtitle) +
      ggplot2::theme(plot.title = element_text(hjust = 0.5, face = "bold",
                                               margin = ggplot2::margin(3, 3, 5, 3)),
                     plot.subtitle = ggtext::element_markdown(hjust = 0.5,
                                                              size = 10))
    
    wc = ggplot2::ggplot(fold_change[gs_content, ],
                         aes(label = gene_name, size = abs(metric), color = metric)) +
      ggwordcloud::geom_text_wordcloud_area(show.legend = TRUE) +
      ggplot2::scale_color_gradient2(
        name = metric,
        low = aquarius::color_cnv[1],
        mid = "gray70", midpoint = 0,
        high = aquarius::color_cnv[3]) +
      ggplot2::scale_size_area(max_size = 7) +
      ggplot2::theme_minimal() +
      ggplot2::guides(size = "none")
    
    return(list(p, wc))
  }) %>% unlist(., recursive = FALSE)
  
  return(plot_list)
}

Visualization

Gene expression

We visualize gene expression for some markers :

features = c("percent.mt", "percent.rb", "nFeature_RNA")

plot_list = lapply(features, FUN = function(one_gene) {
  Seurat::FeaturePlot(sobj, features = one_gene,
                      reduction = name2D) +
    ggplot2::theme(aspect.ratio = 1) +
    ggplot2::scale_color_gradientn(colors = aquarius::color_gene) +
    Seurat::NoAxes()
})

patchwork::wrap_plots(plot_list, ncol = 3)

Clusters

We visualize clusters :

cluster_plot = Seurat::DimPlot(sobj, reduction = name2D, label = TRUE) +
  Seurat::NoAxes() +
  ggplot2::theme(aspect.ratio = 1)
cluster_plot

Cell type

We visualize cell type split by sample :

plot_list = aquarius::plot_split_dimred(sobj,
                                        reduction = name2D,
                                        split_by = "project_name",
                                        group_by = "cell_type",
                                        split_color = setNames(sample_info$color,
                                                               nm = sample_info$project_name),
                                        group_color = color_markers,
                                        bg_pt_size = 0.5, main_pt_size = 0.5)

plot_list[[length(plot_list) + 1]] = cluster_plot

patchwork::wrap_plots(plot_list, ncol = 4) &
  Seurat::NoLegend()

Cluster type

We summarize major cell type by cluster :

cell_type_clusters = sobj@meta.data[, c("cell_type", "seurat_clusters")] %>%
  table() %>%
  prop.table(., margin = 2) %>%
  apply(., 2, which.max)
cell_type_clusters = setNames(levels(sobj$cell_type)[cell_type_clusters],
                              nm = names(cell_type_clusters))

We define cluster type :

sobj$cluster_type = cell_type_clusters[sobj$seurat_clusters] %>%
  as.factor()
table(sobj$cluster_type, sobj$cell_type)

##                   
##                    CD4 T cells CD8 T cells Langerhans cells macrophages B cells
##   B cells                    0           0                0           0      30
##   CD4 T cells              769          64                2           2       2
##   CD8 T cells               91         533                1           0       5
##   Langerhans cells          22           0              257          23       8
##   macrophages                2           0                7         450       1
##   proliferative              0           0               15           0       0
##                   
##                    cuticle cortex medulla IRS proliferative IBL ORS IFE HFSC
##   B cells                0      0       0   0             0   0   0   0    0
##   CD4 T cells            2      0       3   3             0   1   0   0    0
##   CD8 T cells            0      0       0   0             0   0   0   0    0
##   Langerhans cells       1      1       1   3             3   0   0   0    1
##   macrophages            0      0       0   0             0   2   0   0    0
##   proliferative          0      0       0   0            22   0   0   0    0
##                   
##                    sebocytes
##   B cells                  0
##   CD4 T cells              0
##   CD8 T cells              0
##   Langerhans cells         1
##   macrophages              1
##   proliferative            0

We subset color_markers :

color_markers = color_markers[levels(sobj$cluster_type)]

We compare cluster annotation and cell type annotation :

p1 = Seurat::DimPlot(sobj, group.by = "cell_type",
                     reduction = name2D, cols = color_markers) +
  ggplot2::labs(title = "Cell type") +
  Seurat::NoAxes() +
  ggplot2::theme(aspect.ratio = 1,
                 plot.title = element_text(hjust = 0.5))

p2 = Seurat::DimPlot(sobj, group.by = "cluster_type",
                     reduction = name2D, cols = color_markers) +
  ggplot2::labs(title = "Cluster type") +
  Seurat::NoAxes() +
  ggplot2::theme(aspect.ratio = 1,
                 plot.title = element_text(hjust = 0.5))

patchwork::wrap_plots(p1, p2, guides = "collect")

Differential expression

For each population, what are the differences between healthy donors (HD) and HS patients (HS) ? We save the results in a list :

list_results = list()

We make over-representation analysis for each group of genes. We load gene sets from MSigDB :

gene_sets = aquarius::get_gene_sets(species = "Homo sapiens")
gene_sets = gene_sets$gene_sets

head(gene_sets)

## # A tibble: 6 x 16
##   gs_cat gs_subcat gs_name gene_symbol entrez_gene ensembl_gene human_gene_symb~
##   <chr>  <chr>     <chr>   <chr>             <int> <chr>        <chr>           
## 1 C5     GO:BP     GOBP_1~ AASDHPPT          60496 ENSG0000014~ AASDHPPT        
## 2 C5     GO:BP     GOBP_1~ ALDH1L1           10840 ENSG0000014~ ALDH1L1         
## 3 C5     GO:BP     GOBP_1~ ALDH1L2          160428 ENSG0000013~ ALDH1L2         
## 4 C5     GO:BP     GOBP_1~ MTHFD1             4522 ENSG0000010~ MTHFD1          
## 5 C5     GO:BP     GOBP_1~ MTHFD1L           25902 ENSG0000012~ MTHFD1L         
## 6 C5     GO:BP     GOBP_1~ MTHFD2L          441024 ENSG0000016~ MTHFD2L         
## # ... with 9 more variables: human_entrez_gene <int>, human_ensembl_gene <chr>,
## #   gs_id <chr>, gs_pmid <chr>, gs_geoid <chr>, gs_exact_source <chr>,
## #   gs_url <chr>, gs_description <chr>, category <chr>

How many gene sets ?

gene_sets[, c("gs_subcat", "gs_name")] %>%
  dplyr::distinct() %>%
  dplyr::pull(gs_subcat) %>%
  table() %>%
  as.data.frame.table() %>%
  `colnames<-`(c("Category", "Nb gene sets"))

##          Category Nb gene sets
## 1                           50
## 2         CP:KEGG          186
## 3          CP:PID          196
## 4     CP:REACTOME         1615
## 5 CP:WIKIPATHWAYS          664
## 6           GO:BP         7658
## 7           GO:CC         1006
## 8           GO:MF         1738

We get gene name and gene ID correspondence :

gene_corresp = sobj@assays[["RNA"]]@meta.features[, c("gene_name", "Ensembl_ID")] %>%
  `colnames<-`(c("NAME", "ID")) %>%
  dplyr::mutate(ID = as.character(ID))
rownames(gene_corresp) = gene_corresp$ID

head(gene_corresp)

##                       NAME              ID
## ENSG00000238009 AL627309.1 ENSG00000238009
## ENSG00000237491 AL669831.5 ENSG00000237491
## ENSG00000225880  LINC00115 ENSG00000225880
## ENSG00000230368     FAM41C ENSG00000230368
## ENSG00000187634     SAMD11 ENSG00000187634
## ENSG00000188976      NOC2L ENSG00000188976

Langerhans cells

group_name = "Langerhans cells"

Langerhans cells are those clusters :

clusters_oi = names(which(cell_type_clusters == "Langerhans cells"))
clusters_oi

## [1] "3"  "7"  "8"  "12"

We represent those clusters on the projection :

see_clusters("Langerhans cells")

DE

We perform differential expression between HS and HD, within this population :

subsobj = subset(sobj, seurat_clusters %in% c(3, 8))
Seurat::Idents(subsobj) = subsobj$sample_type

table(subsobj$sample_type)

## 
##  HS  HD 
## 148  57

We make identify specific markers for these group :

mark = Seurat::FindMarkers(subsobj, ident.1 = "HS", ident.2 = "HD")

mark = mark %>%
  dplyr::filter(p_val_adj < 0.05) %>%
  dplyr::arrange(-avg_logFC, pct.1 - pct.2)

list_results[[group_name]]$mark = mark

dim(mark)

## [1] 12  5

head(mark, n = 20)

##                 p_val  avg_logFC pct.1 pct.2    p_val_adj
## RPS28    1.761567e-07  0.6392986 0.919 0.842 2.663665e-03
## TMSB10   1.819343e-06  0.5402961 0.966 0.877 2.751029e-02
## HLA-DQA1 2.234450e-07 -0.4536573 0.932 0.947 3.378712e-03
## ZNF302   1.750968e-06 -0.6587976 0.291 0.632 2.647639e-02
## NCALD    3.027328e-07 -0.6827688 0.115 0.439 4.577622e-03
## WDR66    1.286711e-09 -0.8234062 0.176 0.596 1.945635e-05
## GPM6A    2.527528e-07 -0.8876557 0.128 0.456 3.821875e-03
## PER1     2.498950e-06 -1.0169989 0.122 0.386 3.778662e-02
## JUN      1.994633e-07 -1.1092696 0.466 0.737 3.016084e-03
## ACTR3C   5.761218e-07 -1.1418367 0.061 0.316 8.711537e-03
## ZFP36L2  6.359128e-07 -1.1479119 0.486 0.702 9.615638e-03
## TSC22D3  5.439881e-09 -1.2658959 0.426 0.737 8.225644e-05

Up in HS

What are the genes up-regulated in HS ?

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC > 0) %>%
  rownames()
genes_of_interest

## [1] "RPS28"  "TMSB10"

Up in HD

What are the genes up-regulated in HD ?

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC < 0) %>%
  rownames()
genes_of_interest

##  [1] "HLA-DQA1" "ZNF302"   "NCALD"    "WDR66"    "GPM6A"    "PER1"    
##  [7] "JUN"      "ACTR3C"   "ZFP36L2"  "TSC22D3"

GSEA

We run a GSEA for all gene sets, from the full count matrix :

ranked_gene_list = aquarius::run_foldchange(Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts"),
                                            group1 = colnames(subsobj)[subsobj@active.ident %in% "HS"],
                                            group2 = colnames(subsobj)[subsobj@active.ident %in% "HD"])
names(ranked_gene_list) = gene_corresp$ID

gsea_results = aquarius::gsea_run(ranked_gene_list = ranked_gene_list,
                                  gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                  GSEA_p_val_thresh = 1)

list_results[[group_name]]$gsea = gsea_results

gsea_results@result %>%
  dplyr::filter(pvalue < 0.05) %>%
  dplyr::top_n(., n = 200, wt = abs(NES)) %>%
  dplyr::mutate(too_long = ifelse(nchar(ID) > 60, yes = TRUE, no = FALSE)) %>%
  dplyr::mutate(ID = stringr::str_sub(ID, end = 60)) %>%
  dplyr::mutate(ID = ifelse(too_long, yes = paste0(ID, "..."), no = ID)) %>%
  aquarius::gsea_plot(show_legend = TRUE) +
  ggplot2::labs(title = "GSEA using all genes (count matrix)") +
  ggplot2::theme(plot.title = element_text(size = 20))

We compute a fold change table to make a wordcloud :

fold_change = gene_corresp
colnames(fold_change) = c("gene_name", "ID")
rownames(fold_change) = fold_change$ID
fold_change$FC = ranked_gene_list[rownames(fold_change)]
fold_change$pct.1 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HS"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$pct.2 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HD"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$FC_x_pct = ifelse(fold_change$FC > 0,
                              yes = fold_change$FC * fold_change$pct.1,
                              no = fold_change$FC * fold_change$pct.2)

dim(fold_change) ; head(fold_change)

## [1] 15121     6

##                  gene_name              ID         FC      pct.1     pct.2
## ENSG00000238009 AL627309.1 ENSG00000238009 -1.8876775 0.00000000 0.0000000
## ENSG00000237491 AL669831.5 ENSG00000237491 -2.4022507 0.04054054 0.1052632
## ENSG00000225880  LINC00115 ENSG00000225880 -1.3731043 0.05405405 0.0877193
## ENSG00000230368     FAM41C ENSG00000230368  1.2822475 0.05405405 0.0000000
## ENSG00000187634     SAMD11 ENSG00000187634 -1.8876775 0.00000000 0.0000000
## ENSG00000188976      NOC2L ENSG00000188976 -0.7807623 0.16216216 0.1754386
##                    FC_x_pct
## ENSG00000238009  0.00000000
## ENSG00000237491 -0.25286849
## ENSG00000225880 -0.12044775
## ENSG00000230368  0.06931068
## ENSG00000187634  0.00000000
## ENSG00000188976 -0.13697584

We make the gsea plot for some gene sets :

gs_oi = c("GOBP_ANTIGEN_PROCESSING_AND_PRESENTATION_VIA_MHC_CLASS_IB",
          "GOCC_MHC_CLASS_I_PROTEIN_COMPLEX",
          "GOBP_REGULATION_OF_DENDRITIC_CELL_DIFFERENTIATION",
          "GOBP_XENOPHAGY",
          "REACTOME_NCAM1_INTERACTIONS",
          "GOCC_T_CELL_RECEPTOR_COMPLEX")

plot_list = make_gsea_plot(gsea_results, gs_oi, fold_change, "FC_x_pct")

patchwork::wrap_plots(plot_list, ncol = 2, widths = c(2,1.5))

Macrophages

group_name = "macrophages"

Macrophages are those clusters :

clusters_oi = names(which(cell_type_clusters == "macrophages"))
clusters_oi

## [1] "2"  "11"

We represent those clusters on the projection :

see_clusters("macrophages")

DE

We perform differential expression between HS and HD, within this population :

subsobj = subset(sobj, seurat_clusters %in% c(2))
Seurat::Idents(subsobj) = subsobj$sample_type

table(subsobj$sample_type)

## 
##  HS  HD 
## 378  31

We make identify specific markers for these group :

mark = Seurat::FindMarkers(subsobj, ident.1 = "HS", ident.2 = "HD")

mark = mark %>%
  dplyr::filter(p_val_adj < 0.05) %>%
  dplyr::arrange(-avg_logFC, pct.1 - pct.2)

list_results[[group_name]]$mark = mark

dim(mark)

## [1] 47  5

head(mark, n = 20)

##                  p_val  avg_logFC pct.1 pct.2    p_val_adj
## HLA-DRB5  1.631166e-14  1.6933362 0.963 0.226 2.466487e-10
## HLA-C     1.656456e-18  1.6046388 0.995 0.710 2.504727e-14
## MTRNR2L12 5.065078e-11  0.9147492 0.989 1.000 7.658905e-07
## RPS26     1.257519e-06  0.7792169 0.995 0.968 1.901495e-02
## MTRNR2L10 1.748459e-07  0.7368945 0.804 0.355 2.643845e-03
## MTRNR2L8  3.214200e-07  0.7163832 0.968 0.871 4.860192e-03
## HLA-DQA2  1.009694e-07  0.7077588 0.772 0.194 1.526759e-03
## C1QC      4.803251e-11  0.6511403 1.000 1.000 7.262995e-07
## C1QA      3.100723e-11  0.6175805 1.000 0.968 4.688603e-07
## B2M       5.202266e-14  0.5888569 1.000 1.000 7.866346e-10
## TMSB4X    1.177656e-13  0.5886220 1.000 1.000 1.780733e-09
## HLA-A     8.822978e-09  0.5814968 1.000 1.000 1.334123e-04
## C1QB      1.053374e-08  0.5424144 0.997 1.000 1.592806e-04
## HLA-DPA1  1.267576e-06  0.3537871 1.000 1.000 1.916701e-02
## RPL23A    2.907519e-06  0.2998397 1.000 1.000 4.396459e-02
## CHST13    1.541856e-09 -0.2636116 0.013 0.194 2.331440e-05
## RPS29     4.606266e-08 -0.3696506 0.997 1.000 6.965134e-04
## NCALD     1.703572e-08 -0.3733028 0.098 0.452 2.575971e-04
## MT-CYB    3.008289e-07 -0.3740895 0.989 1.000 4.548834e-03
## VEGFA     2.205237e-07 -0.4069963 0.079 0.355 3.334539e-03

Up in HS

We explore enrichment in gene sets for HS population.

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC > 0) %>%
  rownames()

enrichr_results = aquarius::run_enrichr(gene_names = genes_of_interest,
                                        gene_corresp = gene_corresp,
                                        gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                        make_plot = TRUE,
                                        plot_title = "Up-regulated in HS compared to HD")

list_results[[group_name]]$enrichr_hs = enrichr_results$ego

enrichr_results$plot +
  ggplot2::theme(axis.text.y = element_text(size = 8))

Up in HD

We explore enrichment in gene sets for HD population.

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC < 0) %>%
  rownames()

enrichr_results = aquarius::run_enrichr(gene_names = genes_of_interest,
                                        gene_corresp = gene_corresp,
                                        gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                        make_plot = TRUE,
                                        plot_title = "Down-regulated in HS compared to HD")

list_results[[group_name]]$enrichr_hd = enrichr_results$ego

enrichr_results$plot +
  ggplot2::theme(axis.text.y = element_text(size = 8))

GSEA

We run a GSEA for all gene sets, from the full count matrix :

ranked_gene_list = aquarius::run_foldchange(Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts"),
                                            group1 = colnames(subsobj)[subsobj@active.ident %in% "HS"],
                                            group2 = colnames(subsobj)[subsobj@active.ident %in% "HD"])
names(ranked_gene_list) = gene_corresp$ID

gsea_results = aquarius::gsea_run(ranked_gene_list = ranked_gene_list,
                                  gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                  GSEA_p_val_thresh = 1)

list_results[[group_name]]$gsea = gsea_results

gsea_results@result %>%
  dplyr::filter(pvalue < 0.05) %>%
  dplyr::top_n(., n = 200, wt = abs(NES)) %>%
  dplyr::mutate(too_long = ifelse(nchar(ID) > 70, yes = TRUE, no = FALSE)) %>%
  dplyr::mutate(ID = stringr::str_sub(ID, end = 70)) %>%
  dplyr::mutate(ID = ifelse(too_long, yes = paste0(ID, "..."), no = ID)) %>%
  aquarius::gsea_plot(show_legend = TRUE) +
  ggplot2::labs(title = "GSEA using all genes (count matrix)") +
  ggplot2::theme(plot.title = element_text(size = 20))

We compute a fold change table to make a wordcloud :

fold_change = gene_corresp
colnames(fold_change) = c("gene_name", "ID")
rownames(fold_change) = fold_change$ID
fold_change$FC = ranked_gene_list[rownames(fold_change)]
fold_change$pct.1 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HS"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$pct.2 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HD"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$FC_x_pct = ifelse(fold_change$FC > 0,
                              yes = fold_change$FC * fold_change$pct.1,
                              no = fold_change$FC * fold_change$pct.2)

dim(fold_change) ; head(fold_change)

## [1] 15121     6

##                  gene_name              ID          FC       pct.1      pct.2
## ENSG00000238009 AL627309.1 ENSG00000238009 -3.06769505 0.002645503 0.00000000
## ENSG00000237491 AL669831.5 ENSG00000237491 -0.48273255 0.145502646 0.12903226
## ENSG00000225880  LINC00115 ENSG00000225880 -0.06769505 0.076719577 0.03225806
## ENSG00000230368     FAM41C ENSG00000230368  1.51726745 0.113756614 0.00000000
## ENSG00000187634     SAMD11 ENSG00000187634 -3.06769505 0.002645503 0.00000000
## ENSG00000188976      NOC2L ENSG00000188976  0.39173657 0.283068783 0.16129032
##                     FC_x_pct
## ENSG00000238009  0.000000000
## ENSG00000237491 -0.062288070
## ENSG00000225880 -0.002183711
## ENSG00000230368  0.172599208
## ENSG00000187634  0.000000000
## ENSG00000188976  0.110888395

We make the gsea plot for some gene sets :

gs_oi = c("GOBP_B_CELL_MEDIATED_IMMUNITY",
          "REACTOME_INTERFERON_ALPHA_BETA_SIGNALING",
          "GOCC_MHC_PROTEIN_COMPLEX",
          "GOMF_BETA_2_MICROGLOBULIN_BINDING",
          "GOBP_REGULATION_OF_T_CELL_MEDIATED_IMMUNE_RESPONSE_TO_TUMOR_CELL",
          "GOMF_MHC_CLASS_IB_RECEPTOR_ACTIVITY",
          "REACTOME_NCAM1_INTERACTIONS",
          "GOBP_NEURON_FATE_SPECIFICATION")

plot_list = make_gsea_plot(gsea_results, gs_oi, fold_change, "FC_x_pct")

patchwork::wrap_plots(plot_list, ncol = 2, widths = c(2,1.5))

CD4 T cells

group_name = "CD4 T cells"

CD4 T cells are those clusters :

clusters_oi = names(which(cell_type_clusters == "CD4 T cells"))
clusters_oi

## [1] "0"  "4"  "9"  "10"

We represent those clusters on the projection :

see_clusters("CD4 T cells")

DE

We perform differential expression between HS and HD, within cluster 0 only, because other clusters contain too few cells and have distinct expression profile from cluster 0 :

subsobj = subset(sobj, seurat_clusters %in% c(0))
Seurat::Idents(subsobj) = subsobj$sample_type

table(subsobj$sample_type)

## 
##  HS  HD 
## 524  79

We make identify specific markers for these group :

mark = Seurat::FindMarkers(subsobj, ident.1 = "HS", ident.2 = "HD")

mark = mark %>%
  dplyr::filter(p_val_adj < 0.05) %>%
  dplyr::arrange(-avg_logFC, pct.1 - pct.2)

list_results[[group_name]]$mark = mark

dim(mark)

## [1] 90  5

head(mark, n = 20)

##                 p_val avg_logFC pct.1 pct.2    p_val_adj
## GZMA     1.964994e-19 1.6362221 0.750 0.253 2.971268e-15
## RPS26    1.098328e-30 1.1912840 0.992 0.835 1.660782e-26
## HLA-C    3.163130e-20 0.9575722 0.952 0.696 4.782969e-16
## KLRB1    6.290488e-08 0.6979863 0.760 0.418 9.511846e-04
## ABRACL   2.349188e-07 0.5919547 0.651 0.392 3.552208e-03
## STK17A   1.597330e-09 0.5617681 0.775 0.494 2.415322e-05
## CLEC2B   6.187239e-07 0.5504522 0.443 0.127 9.355724e-03
## CD3D     6.816875e-10 0.4786997 0.979 0.924 1.030780e-05
## NDUFS5   6.719076e-07 0.4705508 0.718 0.519 1.015991e-02
## CD2      1.358823e-08 0.4418964 0.969 0.911 2.054676e-04
## CYBA     1.168838e-07 0.4097895 0.931 0.848 1.767399e-03
## ARPC3    3.609123e-07 0.4093346 0.882 0.772 5.457355e-03
## SH3BGRL3 1.347683e-10 0.3818926 0.996 0.962 2.037832e-06
## SERF2    1.473128e-08 0.3801104 0.969 0.899 2.227516e-04
## TMSB4X   5.977030e-11 0.3493581 1.000 1.000 9.037868e-07
## H3F3A    2.522627e-07 0.3272609 0.983 0.949 3.814464e-03
## CFL1     8.256907e-08 0.3240998 0.979 0.975 1.248527e-03
## B2M      4.670220e-17 0.3124388 1.000 1.000 7.061839e-13
## TPT1     9.984271e-14 0.3056347 1.000 1.000 1.509722e-09
## RPS27    5.899066e-13 0.3037666 1.000 1.000 8.919978e-09

Up in HS

We explore enrichment in gene sets for HS population.

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC > 0) %>%
  rownames()

enrichr_results = aquarius::run_enrichr(gene_names = genes_of_interest,
                                        gene_corresp = gene_corresp,
                                        gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                        make_plot = TRUE,
                                        plot_title = "Up-regulated in HS compared to HD")

list_results[[group_name]]$enrichr_hs = enrichr_results$ego

enrichr_results$plot +
  ggplot2::theme(axis.text.y = element_text(size = 8))

Up in HD

We explore enrichment in gene sets for HD population.

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC < 0) %>%
  rownames()

enrichr_results = aquarius::run_enrichr(gene_names = genes_of_interest,
                                        gene_corresp = gene_corresp,
                                        gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                        make_plot = TRUE,
                                        plot_title = "Down-regulated in HS compared to HD")

list_results[[group_name]]$enrichr_hd = enrichr_results$ego

enrichr_results$plot +
  ggplot2::theme(axis.text.y = element_text(size = 8))

GSEA

We run a GSEA for all gene sets, from the full count matrix :

ranked_gene_list = aquarius::run_foldchange(Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts"),
                                            group1 = colnames(subsobj)[subsobj@active.ident %in% "HS"],
                                            group2 = colnames(subsobj)[subsobj@active.ident %in% "HD"])
names(ranked_gene_list) = gene_corresp$ID

gsea_results = aquarius::gsea_run(ranked_gene_list = ranked_gene_list,
                                  gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                  GSEA_p_val_thresh = 1)

list_results[[group_name]]$gsea = gsea_results

gsea_results@result %>%
  dplyr::filter(pvalue < 0.05) %>%
  dplyr::top_n(., n = 200, wt = abs(NES)) %>%
  dplyr::mutate(too_long = ifelse(nchar(ID) > 60, yes = TRUE, no = FALSE)) %>%
  dplyr::mutate(ID = stringr::str_sub(ID, end = 60)) %>%
  dplyr::mutate(ID = ifelse(too_long, yes = paste0(ID, "..."), no = ID)) %>%
  aquarius::gsea_plot(show_legend = TRUE) +
  ggplot2::labs(title = "GSEA using all genes (count matrix)") +
  ggplot2::theme(plot.title = element_text(size = 20))

We compute a fold change table to make a wordcloud :

fold_change = gene_corresp
colnames(fold_change) = c("gene_name", "ID")
rownames(fold_change) = fold_change$ID
fold_change$FC = ranked_gene_list[rownames(fold_change)]
fold_change$pct.1 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HS"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$pct.2 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HD"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$FC_x_pct = ifelse(fold_change$FC > 0,
                              yes = fold_change$FC * fold_change$pct.1,
                              no = fold_change$FC * fold_change$pct.2)

dim(fold_change) ; head(fold_change)

## [1] 15121     6

##                  gene_name              ID         FC       pct.1      pct.2
## ENSG00000238009 AL627309.1 ENSG00000238009 -1.1032762 0.003816794 0.00000000
## ENSG00000237491 AL669831.5 ENSG00000237491 -1.3097270 0.022900763 0.05063291
## ENSG00000225880  LINC00115 ENSG00000225880 -1.5507351 0.015267176 0.05063291
## ENSG00000230368     FAM41C ENSG00000230368 -1.6882387 0.001908397 0.00000000
## ENSG00000187634     SAMD11 ENSG00000187634 -2.6882387 0.000000000 0.00000000
## ENSG00000188976      NOC2L ENSG00000188976  0.4615085 0.112595420 0.07594937
##                    FC_x_pct
## ENSG00000238009  0.00000000
## ENSG00000237491 -0.06631529
## ENSG00000225880 -0.07851823
## ENSG00000230368  0.00000000
## ENSG00000187634  0.00000000
## ENSG00000188976  0.05196374

We make the gsea plot for some gene sets :

gs_oi = c("REACTOME_IMMUNOREGULATORY_INTERACTIONS_BETWEEN_A_LYMPHOID_AND_A_NON_LYMPHOID_CELL",
          "PID_RAC1_PATHWAY",
          "REACTOME_PD_1_SIGNALING",
          "WP_CANCER_IMMUNOTHERAPY_BY_PD1_BLOCKADE",
          "GOCC_CYTOLYTIC_GRANULE",
          "GOBP_GRANZYME_MEDIATED_PROGRAMMED_CELL_DEATH_SIGNALING_PATHWAY",
          "REACTOME_INTERLEUKIN_9_SIGNALING",
          "GOMF_INHIBITORY_MHC_CLASS_I_RECEPTOR_ACTIVITY",
          "REACTOME_COMPLEMENT_CASCADE",
          "HALLMARK_TNFA_SIGNALING_VIA_NFKB",
          "WP_NEUROINFLAMMATION")

plot_list = make_gsea_plot(gsea_results, gs_oi, fold_change, "FC_x_pct")

patchwork::wrap_plots(plot_list, ncol = 2, widths = c(2,1.5))

CD8 T cells

group_name = "CD8 T cells"

CD8 T cells are those clusters :

clusters_oi = names(which(cell_type_clusters == "CD8 T cells"))
clusters_oi

## [1] "1" "5" "6"

We represent those clusters on the projection :

see_clusters("CD8 T cells")

DE

We perform differential expression between HS and HD, within cluster 1 only, because other clusters contain too few cells and have distinct expression profile from cluster 1 :

subsobj = subset(sobj, seurat_clusters %in% c(1))
Seurat::Idents(subsobj) = subsobj$sample_type

table(subsobj$sample_type)

## 
##  HS  HD 
## 361  82

We make identify specific markers for these group :

mark = Seurat::FindMarkers(subsobj, ident.1 = "HS", ident.2 = "HD")

mark = mark %>%
  dplyr::filter(p_val_adj < 0.05) %>%
  dplyr::arrange(-avg_logFC, pct.1 - pct.2)

list_results[[group_name]]$mark = mark

dim(mark)

## [1] 61  5

head(mark, n = 20)

##                  p_val avg_logFC pct.1 pct.2    p_val_adj
## RPS26     3.309407e-38 1.4312432 0.986 0.866 5.004155e-34
## HLA-C     4.952396e-25 1.1772438 0.936 0.549 7.488517e-21
## KLRC1     2.997716e-08 0.8124008 0.388 0.061 4.532846e-04
## TMEM154   2.827158e-07 0.7974845 0.338 0.061 4.274946e-03
## KLRB1     5.141186e-19 0.7927007 0.961 0.927 7.773987e-15
## GBP5      3.845596e-08 0.7712944 0.460 0.134 5.814926e-04
## LINC02195 8.195677e-07 0.7136043 0.260 0.012 1.239268e-02
## CTSW      1.852837e-09 0.6638753 0.820 0.549 2.801674e-05
## CD8A      7.446961e-07 0.6525479 0.380 0.085 1.126055e-02
## EPHB6     1.602297e-06 0.6521750 0.518 0.232 2.422833e-02
## RPS27L    1.901202e-06 0.6374235 0.593 0.317 2.874808e-02
## GZMA      1.964021e-06 0.6100182 0.740 0.488 2.969796e-02
## DYNLL1    1.271236e-07 0.5895053 0.676 0.378 1.922236e-03
## MTRNR2L12 1.277002e-07 0.5500563 0.986 0.963 1.930954e-03
## RPS19     6.390492e-11 0.4413521 0.997 1.000 9.663063e-07
## ACTG1     1.365204e-08 0.4375325 0.992 0.951 2.064324e-04
## RPS27     6.308878e-16 0.3647799 1.000 1.000 9.539655e-12
## HLA-A     4.398120e-07 0.3026910 0.997 1.000 6.650397e-03
## ACTB      1.195282e-06 0.2961452 0.997 1.000 1.807385e-02
## B2M       3.195644e-15 0.2808143 1.000 1.000 4.832133e-11

Up in HS

We explore enrichment in gene sets for HS population.

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC > 0) %>%
  rownames()

enrichr_results = aquarius::run_enrichr(gene_names = genes_of_interest,
                                        gene_corresp = gene_corresp,
                                        gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                        make_plot = TRUE,
                                        plot_title = "Up-regulated in HS compared to HD")

list_results[[group_name]]$enrichr_hs = enrichr_results$ego

enrichr_results$plot +
  ggplot2::theme(axis.text.y = element_text(size = 8))

Up in HD

We explore enrichment in gene sets for HD population.

genes_of_interest = mark %>%
  dplyr::filter(avg_logFC < 0) %>%
  rownames()

enrichr_results = aquarius::run_enrichr(gene_names = genes_of_interest,
                                        gene_corresp = gene_corresp,
                                        gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                        make_plot = TRUE,
                                        plot_title = "Down-regulated in HS compared to HD")

list_results[[group_name]]$enrichr_hd = enrichr_results$ego

enrichr_results$plot +
  ggplot2::theme(axis.text.y = element_text(size = 8))

GSEA

We run a GSEA for all gene sets, from the full count matrix :

ranked_gene_list = aquarius::run_foldchange(Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts"),
                                            group1 = colnames(subsobj)[subsobj@active.ident %in% "HS"],
                                            group2 = colnames(subsobj)[subsobj@active.ident %in% "HD"])
names(ranked_gene_list) = gene_corresp$ID

gsea_results = aquarius::gsea_run(ranked_gene_list = ranked_gene_list,
                                  gene_sets = gene_sets[, c("gs_name", "ensembl_gene")],
                                  GSEA_p_val_thresh = 1)

list_results[[group_name]]$gsea = gsea_results

gsea_results@result %>%
  dplyr::filter(pvalue < 0.05) %>%
  dplyr::top_n(., n = 200, wt = abs(NES)) %>%
  dplyr::mutate(too_long = ifelse(nchar(ID) > 60, yes = TRUE, no = FALSE)) %>%
  dplyr::mutate(ID = stringr::str_sub(ID, end = 60)) %>%
  dplyr::mutate(ID = ifelse(too_long, yes = paste0(ID, "..."), no = ID)) %>%
  aquarius::gsea_plot(show_legend = TRUE) +
  ggplot2::labs(title = "GSEA using all genes (count matrix)") +
  ggplot2::theme(plot.title = element_text(size = 20))

We compute a fold change table to make a wordcloud :

fold_change = gene_corresp
colnames(fold_change) = c("gene_name", "ID")
rownames(fold_change) = fold_change$ID
fold_change$FC = ranked_gene_list[rownames(fold_change)]
fold_change$pct.1 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HS"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$pct.2 = Seurat::GetAssayData(subsobj, assay = "RNA", slot = "counts")[, subsobj@active.ident == "HD"] %>%
  apply(., MARGIN = 1, FUN = function(one_row) {
    return( mean(one_row != 0) )
  })
fold_change$FC_x_pct = ifelse(fold_change$FC > 0,
                              yes = fold_change$FC * fold_change$pct.1,
                              no = fold_change$FC * fold_change$pct.2)

dim(fold_change) ; head(fold_change)

## [1] 15121     6

##                  gene_name              ID         FC       pct.1      pct.2
## ENSG00000238009 AL627309.1 ENSG00000238009 -0.6062212 0.005540166 0.00000000
## ENSG00000237491 AL669831.5 ENSG00000237491  0.1307444 0.022160665 0.01219512
## ENSG00000225880  LINC00115 ENSG00000225880 -1.3431868 0.022160665 0.04878049
## ENSG00000230368     FAM41C ENSG00000230368 -0.1911837 0.008310249 0.00000000
## ENSG00000187634     SAMD11 ENSG00000187634 -1.4542181 0.011080332 0.02439024
## ENSG00000188976      NOC2L ENSG00000188976 -0.7317521 0.085872576 0.13414634
##                     FC_x_pct
## ENSG00000238009  0.000000000
## ENSG00000237491  0.002897383
## ENSG00000225880 -0.065521307
## ENSG00000230368  0.000000000
## ENSG00000187634 -0.035468734
## ENSG00000188976 -0.098161863

We make the gsea plot for some gene sets :

gs_oi = c("REACTOME_IMMUNOREGULATORY_INTERACTIONS_BETWEEN_A_LYMPHOID_AND_A_NON_LYMPHOID_CELL",
          "PID_IL12_2PATHWAY",
          "REACTOME_INTERFERON_GAMMA_SIGNALING",
          "GOCC_CYTOLYTIC_GRANULE",
          "GOBP_NATURAL_KILLER_CELL_ACTIVATION_INVOLVED_IN_IMMUNE_RESPONSE",
          "GOBP_ANTIBACTERIAL_HUMORAL_RESPONSE",
          "WP_NEUROINFLAMMATION",
          "HALLMARK_HYPOXIA")

plot_list = make_gsea_plot(gsea_results, gs_oi, fold_change, "FC_x_pct")

patchwork::wrap_plots(plot_list, ncol = 2, widths = c(2,1.5))

Global visualization

We represent differentially expressed genes, only in the four populations of interest :

sobj = subset(sobj, cluster_type %in% c("B cells", "proliferative"), invert = TRUE)
sobj

## An object of class Seurat 
## 15121 features across 2262 samples within 1 assay 
## Active assay: RNA (15121 features, 2000 variable features)
##  6 dimensional reductions calculated: RNA_pca, RNA_pca_20_tsne, RNA_pca_20_umap, harmony, harmony_20_umap, harmony_20_tsne

We extract all DE genes :

features_oi = lapply(list_results, `[[`, 1) %>%
  lapply(., FUN = rownames) %>%
  unlist() %>% unname() %>% unique()

length(features_oi)

## [1] 160

Heatmap

We prepare the scaled expression matrix :

mat_expression = Seurat::GetAssayData(sobj, assay = "RNA", slot = "data")[features_oi, ]
mat_expression = Matrix::t(mat_expression)
mat_expression = dynutils::scale_quantile(mat_expression) # between 0 and 1
mat_expression = Matrix::t(mat_expression)
mat_expression = as.matrix(mat_expression) # not sparse

dim(mat_expression)

## [1]  160 2262

We prepare the heatmap annotation :

ha_top = ComplexHeatmap::HeatmapAnnotation(
  cell_type = sobj$cluster_type,
  sample_type = sobj$sample_type,
  cluster = sobj$seurat_clusters,
  col = list(cell_type = color_markers,
             sample_type = setNames(nm = c("HS", "HD"),
                                    c("#C55F40", "#2C78E6")),
             cluster = setNames(nm = levels(sobj$seurat_clusters),
                                aquarius::gg_color_hue(length(levels(sobj$seurat_clusters))))))

And the heatmap :

sobj$cell_group = paste0(sobj$cluster_type, sobj$sample_type) %>%
  as.factor()

ht = ComplexHeatmap::Heatmap(mat_expression,
                             col = aquarius::color_cnv,
                             # Annotation
                             top_annotation = ha_top,
                             # Grouping
                             column_order = sobj@meta.data %>%
                               dplyr::arrange(cluster_type, sample_type, seurat_clusters) %>%
                               rownames(),
                             column_split = sobj$cell_group,
                             column_gap = rep(unit(c(0.01, 2), "mm"), 4),
                             column_title = NULL,
                             cluster_rows = FALSE,
                             cluster_columns = FALSE,
                             show_column_names = FALSE,
                             # Visual aspect
                             show_heatmap_legend = TRUE,
                             border = TRUE)

ComplexHeatmap::draw(ht,
                     merge_legend = TRUE,
                     heatmap_legend_side = "bottom",
                     annotation_legend_side = "bottom")

Dotplot

Seurat::DotPlot(sobj,
                assay = "RNA",
                features = features_oi,
                group.by = "cell_group") +
  ggplot2::coord_flip() +
  ggplot2::scale_color_gradientn(colors = aquarius::color_gene) +
  ggplot2::theme(axis.title.x = element_blank(),
                 axis.title.y = element_blank(),
                 axis.text.x = element_text(angle = 30, hjust = 1))

Save

We save the list of results :

saveRDS(list_results, file = paste0(out_dir, "/", save_name, "_list_results.rds"))

R Session

show

## R version 3.6.3 (2020-02-29)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 20.04.6 LTS
## 
## Matrix products: default
## BLAS:   /usr/local/lib/R/lib/libRblas.so
## LAPACK: /usr/local/lib/R/lib/libRlapack.so
## 
## locale:
## [1] C
## 
## attached base packages:
##  [1] parallel  stats4    grid      stats     graphics  grDevices utils    
##  [8] datasets  methods   base     
## 
## other attached packages:
##  [1] org.Mm.eg.db_3.10.0   AnnotationDbi_1.48.0  IRanges_2.20.2       
##  [4] S4Vectors_0.24.4      Biobase_2.46.0        BiocGenerics_0.32.0  
##  [7] ComplexHeatmap_2.14.0 ggplot2_3.3.5         patchwork_1.1.2      
## [10] dplyr_1.0.7          
## 
## loaded via a namespace (and not attached):
##   [1] softImpute_1.4              graphlayouts_0.7.0         
##   [3] pbapply_1.4-2               lattice_0.20-41            
##   [5] haven_2.3.1                 vctrs_0.3.8                
##   [7] usethis_2.0.1               dynwrap_1.2.1              
##   [9] blob_1.2.1                  survival_3.2-13            
##  [11] prodlim_2019.11.13          dynutils_1.0.5             
##  [13] later_1.3.0                 DBI_1.1.1                  
##  [15] R.utils_2.11.0              SingleCellExperiment_1.8.0 
##  [17] rappdirs_0.3.3              uwot_0.1.8                 
##  [19] dqrng_0.2.1                 jpeg_0.1-8.1               
##  [21] zlibbioc_1.32.0             pspline_1.0-18             
##  [23] pcaMethods_1.78.0           mvtnorm_1.1-1              
##  [25] htmlwidgets_1.5.4           GlobalOptions_0.1.2        
##  [27] future_1.22.1               UpSetR_1.4.0               
##  [29] laeken_0.5.2                leiden_0.3.3               
##  [31] clustree_0.4.3              scater_1.14.6              
##  [33] irlba_2.3.3                 markdown_1.1               
##  [35] DEoptimR_1.0-9              tidygraph_1.1.2            
##  [37] Rcpp_1.0.9                  readr_2.0.2                
##  [39] KernSmooth_2.23-17          carrier_0.1.0              
##  [41] promises_1.1.0              gdata_2.18.0               
##  [43] DelayedArray_0.12.3         limma_3.42.2               
##  [45] graph_1.64.0                RcppParallel_5.1.4         
##  [47] Hmisc_4.4-0                 fs_1.5.2                   
##  [49] RSpectra_0.16-0             fastmatch_1.1-0            
##  [51] ranger_0.12.1               digest_0.6.25              
##  [53] png_0.1-7                   sctransform_0.2.1          
##  [55] cowplot_1.0.0               DOSE_3.12.0                
##  [57] here_1.0.1                  TInGa_0.0.0.9000           
##  [59] ggraph_2.0.3                pkgconfig_2.0.3            
##  [61] GO.db_3.10.0                DelayedMatrixStats_1.8.0   
##  [63] gower_0.2.1                 ggbeeswarm_0.6.0           
##  [65] iterators_1.0.12            DropletUtils_1.6.1         
##  [67] reticulate_1.26             clusterProfiler_3.14.3     
##  [69] SummarizedExperiment_1.16.1 circlize_0.4.15            
##  [71] beeswarm_0.4.0              GetoptLong_1.0.5           
##  [73] xfun_0.35                   bslib_0.3.1                
##  [75] zoo_1.8-10                  tidyselect_1.1.0           
##  [77] reshape2_1.4.4              purrr_0.3.4                
##  [79] ica_1.0-2                   pcaPP_1.9-73               
##  [81] viridisLite_0.3.0           rtracklayer_1.46.0         
##  [83] rlang_1.0.2                 hexbin_1.28.1              
##  [85] jquerylib_0.1.4             dyneval_0.9.9              
##  [87] glue_1.4.2                  RColorBrewer_1.1-2         
##  [89] matrixStats_0.56.0          stringr_1.4.0              
##  [91] lava_1.6.7                  europepmc_0.3              
##  [93] DESeq2_1.26.0               recipes_0.1.17             
##  [95] labeling_0.3                httpuv_1.5.2               
##  [97] class_7.3-17                BiocNeighbors_1.4.2        
##  [99] DO.db_2.9                   annotate_1.64.0            
## [101] jsonlite_1.7.2              XVector_0.26.0             
## [103] bit_4.0.4                   mime_0.9                   
## [105] aquarius_0.1.5              Rsamtools_2.2.3            
## [107] gridExtra_2.3               gplots_3.0.3               
## [109] stringi_1.4.6               processx_3.5.2             
## [111] gsl_2.1-6                   bitops_1.0-6               
## [113] cli_3.0.1                   batchelor_1.2.4            
## [115] RSQLite_2.2.0               randomForest_4.6-14        
## [117] tidyr_1.1.4                 data.table_1.14.2          
## [119] rstudioapi_0.13             GenomicAlignments_1.22.1   
## [121] nlme_3.1-147                qvalue_2.18.0              
## [123] scran_1.14.6                locfit_1.5-9.4             
## [125] scDblFinder_1.1.8           listenv_0.8.0              
## [127] ggthemes_4.2.4              gridGraphics_0.5-0         
## [129] R.oo_1.24.0                 dbplyr_1.4.4               
## [131] TTR_0.24.2                  readxl_1.3.1               
## [133] lifecycle_1.0.1             timeDate_3043.102          
## [135] ggpattern_0.3.1             munsell_0.5.0              
## [137] cellranger_1.1.0            R.methodsS3_1.8.1          
## [139] proxyC_0.1.5                visNetwork_2.0.9           
## [141] caTools_1.18.0              codetools_0.2-16           
## [143] ggwordcloud_0.5.0           GenomeInfoDb_1.22.1        
## [145] vipor_0.4.5                 lmtest_0.9-38              
## [147] msigdbr_7.5.1               htmlTable_1.13.3           
## [149] triebeard_0.3.0             lsei_1.2-0                 
## [151] xtable_1.8-4                ROCR_1.0-7                 
## [153] BiocManager_1.30.10         scatterplot3d_0.3-41       
## [155] abind_1.4-5                 farver_2.0.3               
## [157] parallelly_1.28.1           RANN_2.6.1                 
## [159] askpass_1.1                 GenomicRanges_1.38.0       
## [161] RcppAnnoy_0.0.16            tibble_3.1.5               
## [163] ggdendro_0.1-20             cluster_2.1.0              
## [165] future.apply_1.5.0          Seurat_3.1.5               
## [167] dendextend_1.15.1           Matrix_1.3-2               
## [169] ellipsis_0.3.2              prettyunits_1.1.1          
## [171] lubridate_1.7.9             ggridges_0.5.2             
## [173] igraph_1.2.5                RcppEigen_0.3.3.7.0        
## [175] fgsea_1.12.0                remotes_2.4.2              
## [177] scBFA_1.0.0                 destiny_3.0.1              
## [179] VIM_6.1.1                   testthat_3.1.0             
## [181] htmltools_0.5.2             BiocFileCache_1.10.2       
## [183] yaml_2.2.1                  utf8_1.1.4                 
## [185] plotly_4.9.2.1              XML_3.99-0.3               
## [187] ModelMetrics_1.2.2.2        e1071_1.7-3                
## [189] foreign_0.8-76              withr_2.5.0                
## [191] fitdistrplus_1.0-14         BiocParallel_1.20.1        
## [193] xgboost_1.4.1.1             bit64_4.0.5                
## [195] foreach_1.5.0               robustbase_0.93-9          
## [197] Biostrings_2.54.0           GOSemSim_2.13.1            
## [199] rsvd_1.0.3                  memoise_2.0.0              
## [201] evaluate_0.18               forcats_0.5.0              
## [203] rio_0.5.16                  geneplotter_1.64.0         
## [205] tzdb_0.1.2                  caret_6.0-86               
## [207] ps_1.6.0                    DiagrammeR_1.0.6.1         
## [209] curl_4.3                    fdrtool_1.2.15             
## [211] fansi_0.4.1                 highr_0.8                  
## [213] urltools_1.7.3              xts_0.12.1                 
## [215] GSEABase_1.48.0             acepack_1.4.1              
## [217] edgeR_3.28.1                checkmate_2.0.0            
## [219] scds_1.2.0                  cachem_1.0.6               
## [221] npsurv_0.4-0                babelgene_22.3             
## [223] rjson_0.2.20                openxlsx_4.1.5             
## [225] ggrepel_0.9.1               clue_0.3-60                
## [227] rprojroot_2.0.2             stabledist_0.7-1           
## [229] tools_3.6.3                 sass_0.4.0                 
## [231] nichenetr_1.1.1             magrittr_2.0.1             
## [233] RCurl_1.98-1.2              proxy_0.4-24               
## [235] car_3.0-11                  ape_5.3                    
## [237] ggplotify_0.0.5             xml2_1.3.2                 
## [239] httr_1.4.2                  assertthat_0.2.1           
## [241] rmarkdown_2.18              boot_1.3-25                
## [243] globals_0.14.0              R6_2.4.1                   
## [245] Rhdf5lib_1.8.0              nnet_7.3-14                
## [247] RcppHNSW_0.2.0              progress_1.2.2             
## [249] genefilter_1.68.0           statmod_1.4.34             
## [251] gtools_3.8.2                shape_1.4.6                
## [253] HDF5Array_1.14.4            BiocSingular_1.2.2         
## [255] rhdf5_2.30.1                splines_3.6.3              
## [257] AUCell_1.8.0                carData_3.0-4              
## [259] colorspace_1.4-1            generics_0.1.0             
## [261] base64enc_0.1-3             dynfeature_1.0.0           
## [263] smoother_1.1                gridtext_0.1.1             
## [265] pillar_1.6.3                tweenr_1.0.1               
## [267] sp_1.4-1                    ggplot.multistats_1.0.0    
## [269] rvcheck_0.1.8               GenomeInfoDbData_1.2.2     
## [271] plyr_1.8.6                  gtable_0.3.0               
## [273] zip_2.2.0                   knitr_1.41                 
## [275] latticeExtra_0.6-29         biomaRt_2.42.1             
## [277] fastmap_1.1.0               ADGofTest_0.3              
## [279] copula_1.0-0                doParallel_1.0.15          
## [281] vcd_1.4-8                   babelwhale_1.0.1           
## [283] openssl_1.4.1               scales_1.1.1               
## [285] backports_1.2.1             ipred_0.9-12               
## [287] enrichplot_1.6.1            hms_1.1.1                  
## [289] ggforce_0.3.1               Rtsne_0.15                 
## [291] shiny_1.7.1                 numDeriv_2016.8-1.1        
## [293] polyclip_1.10-0             lazyeval_0.2.2             
## [295] Formula_1.2-3               tsne_0.1-3                 
## [297] crayon_1.3.4                MASS_7.3-54                
## [299] pROC_1.16.2                 viridis_0.5.1              
## [301] dynparam_1.0.0              rpart_4.1-15               
## [303] zinbwave_1.8.0              compiler_3.6.3             
## [305] ggtext_0.1.0